Re: Build failure with GCC 15 in drivers/gpu/drm/display/drm_dp_dual_mode_helper.c

On Tue, Oct 01, 2024 at 05:02:15PM +0300, Ville Syrjälä wrote:
> On Tue, Oct 01, 2024 at 07:18:23PM +0530, Brahmajit wrote:
> > On 01.10.2024 16:28, Ville Syrjälä wrote:
> > > Looks like gcc is warning about perfectly legal code
> > > (as far as the C standard goes).
> > > 
> > > But we could work around by adding room for the '\0' and
> > > s/sizeof(...)/DP_DUAL_MODE_HDMI_ID_LEN/ in the memcmp().
> > 
> > Sorry I'm new, but can we do something like this?
> > 
> > --- a/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c
> > +++ b/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c
> > @@ -158,13 +158,13 @@ ssize_t drm_dp_dual_mode_write(struct i2c_adapter *adapter,
> >  }
> >  EXPORT_SYMBOL(drm_dp_dual_mode_write);
> >  
> > -static bool is_hdmi_adaptor(const char hdmi_id[DP_DUAL_MODE_HDMI_ID_LEN])
> > +static bool is_hdmi_adaptor(const char hdmi_id[DP_DUAL_MODE_HDMI_ID_LEN+1])
> 
> This should not have the +1
> 
> >  {
> > -	static const char dp_dual_mode_hdmi_id[DP_DUAL_MODE_HDMI_ID_LEN] =

Also we shuld probably remind why the +1 is there.
So a comment here would be good. 

Eg.
/* +1 to avoid spurious -W<whatever> warnings */

> > +	static const char dp_dual_mode_hdmi_id[DP_DUAL_MODE_HDMI_ID_LEN+1] =
> >  		"DP-HDMI ADAPTOR\x04";
> >  
> >  	return memcmp(hdmi_id, dp_dual_mode_hdmi_id,
> > -		      sizeof(dp_dual_mode_hdmi_id)) == 0;
> > +		      DP_DUAL_MODE_HDMI_ID_LEN+1) == 0;
> 
> and neither should this
> 
> >  }
> >  
> >  static bool is_type1_adaptor(uint8_t adaptor_id)
> > 
> > -- 
> > Regards,
> > Brahmajit
> 
> -- 
> Ville Syrjälä
> Intel

-- 
Ville Syrjälä
Intel